EU AI Act Timeline Clarifies for SMBs

What happened: EU regulators published a phased compliance checklist for general-purpose AI providers, and shifted the first hard enforcement deadline for high-risk categories to Q4 2026 — later than the original schedule.

Why it matters: Most small and mid-size teams building on top of foundation models (not training their own) fall under "limited risk" obligations — mainly transparency disclosures — and won't hit the stricter high-risk tier until 2027 unless they're in specifically listed sectors (biometrics, critical infrastructure, employment screening).

Who should care

• Founders building AI features who've been putting off a compliance read
• Teams shipping into regulated sectors (HR, finance, healthcare) inside the EU
• Anyone who assumed the Act applied uniformly to all AI usage

Our take

The phased approach buys time, but "not in scope yet" isn't the same as "never." Worth a calendar reminder for Q3 2026, not a permanent shelf item.

Originally published on ayraix.com, practical AI for enterprise builders.